Why is hashcat only using CPU when I have a GPU?
GPU drivers are not properly installed or detected. Run hashcat -I to see detected devices. For NVIDIA, ensure CUDA Toolkit is installed; for AMD, install ROCm.
32
Hashcat 7.1.2 – Advanced Password Recovery Tool
World's fastest password recovery with GPU acceleration
Hashcat is the world's fastest password cracker supporting GPU acceleration across AMD, NVIDIA, and Intel platforms with 450+ hash algorithms.
About Hashcat
Hashcat represents the pinnacle of password recovery technology, delivering world-record cracking speeds through GPU acceleration. First released in 2008, it has evolved into the industry standard for offline password auditing. Version 7.1.2 provides 223% speed improvement for NetNTLMv2 and 320% acceleration for Scrypt. Hashcat supports 450+ hash algorithms across CPU, GPU, and APU architectures.
System Requirements
- Operating System: Windows 7 SP1 64-bit or later, Ubuntu 18.04+, Fedora 33+, macOS 10.13 or later
- Processor: Intel/AMD 64-bit processor with SSE4.1 or higher
- RAM: 2 GB minimum (4 GB recommended for GPU acceleration)
- Disk Space: 200 MB for installation plus space for wordlists
- Additional Requirements: NVIDIA CUDA Toolkit 11.4+, AMD ROCm 4.2+, Intel GPU runtime, GPU with 2GB+ VRAM
Features Of Hashcat
- GPU-accelerated password cracking with NVIDIA CUDA and AMD ROCm
- 450+ hash algorithms including bcrypt, scrypt, Argon2
- In-kernel rule engine for sophisticated password mutation
- Multi-device cracking with parallel processing across GPUs
- Brute-force, dictionary, hybrid, and mask-based attack modes
- Automatic performance tuning and GPU memory management
- Session support with pause and resume capability
- Distributed cracking across networks for large-scale assessments
- Benchmarking tools for performance profiling
- Custom rule development framework
- Wordlist generation and dictionary management
- Detailed progress reporting and statistics
Pros & Cons
Pros
- Fastest password cracker with 223-320% performance improvements
- Completely free and open-source with no licensing
- Supports 450+ hash algorithms covering enterprise systems
- GPU-accelerated with multi-device support for parallelization
- Industry-unique in-kernel rule engine
- Active development with regular updates
- Massive community with shared rulesets and wordlists
- Python API support for automation frameworks
Cons
- Steep learning curve with complex command-line interface
- Requires expensive GPU hardware for reasonable performance
- GPU driver compatibility issues and version dependencies
- No built-in GUI for visual interaction
- Memory-intensive for large wordlists and complex rules
- Doesn't support online password testing like Hydra
- Required CPU architecture support excludes legacy systems
- GPU memory limitations restrict hash count on smaller devices
Changelog
Version 7.1.2 (August 23, 2025):
- Improved GPU memory allocation for larger hash sets
- Bug fixes for rule engine edge cases
Version 7.1.1 (August 18, 2025):
- Critical stability improvements for long-running sessions
Version 7.1.0 (August 16, 2025):
- Refined autotuning system
- Additional hash algorithm support
Frequently Asked Questions
How do I create custom rules for my organization's password patterns?
Write rules following hashcat rule syntax using operations like c for capitalize, l for lowercase, d for duplicate. Test rules against known password samples first.
What is the most effective attack strategy for unknown password hashes?
Start with dictionary attacks using common wordlists, apply rules, then try mask attacks targeting patterns. Finally use brute-force for short passwords.
How do I integrate hashcat into penetration testing automation?
Use Python hashcat libraries or shell script wrappers to call hashcat with dynamic parameters. Store results in database and correlate with other assessment findings.
Can hashcat crack modern bcrypt passwords in reasonable time?
No. Bcrypt with cost factor 12+ requires 2^12 iterations per guess, making cracking infeasible. Hashcat best targets faster algorithms like MD5 and NTLM.
